> ## Documentation Index
> Fetch the complete documentation index at: https://docs.videngineer.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Scorecard

> A format-aware score: every video judged on the dimensions that matter for its format, with weights and a one-line read per dimension.

The Scorecard answers: **how well does this video execute, for what it is?** A 15-second B2B ad and a 33-minute documentary shouldn't be graded on the same rubric — so videngineer first classifies the format (see [What it's selling](/layers/what-its-selling)), then scores the video on a dimension set specific to that format. Same container, format-specific dimensions and weights.

## The output

| Field        | What it holds                                                                                        |
| ------------ | ---------------------------------------------------------------------------------------------------- |
| `format`     | The format this video was scored as (from the classification layer)                                  |
| `overall`    | 0–10, the weighted mean of the dimension scores                                                      |
| `dimensions` | One entry per dimension: `key`, `label`, `score` (0–10), `read` (a one-line concrete note), `weight` |
| `verdict`    | A one-line synthesis of the whole scorecard                                                          |
| `benchmark`  | Which band the overall score lands in                                                                |

### Benchmark bands

| Overall   | Read                              |
| --------- | --------------------------------- |
| 8.0+      | Top-tier execution for its format |
| 6.5–7.9   | Solid; a few fixable weak points  |
| 5.0–6.4   | Middling — clear room to sharpen  |
| Below 5.0 | Below the bar for this format     |

## Dimension sets by format

Each format gets its own dimensions and weights. The B2B ad rubric, for example:

| Dimension            | Weight | What's judged                                                                  |
| -------------------- | ------ | ------------------------------------------------------------------------------ |
| Hook · first 3s      | 0.30   | Does the open grab with a concrete, countable cost or tension — not a tagline? |
| Problem → cut → calm | 0.25   | Is there a clean escalate → hard-cut → reveal arc? Is the cut decisive?        |
| Brand restraint      | 0.15   | Is the product held back until the calm, with minimal screen time?             |
| Direct CTA           | 0.15   | Is the close direct — brand and URL clear — not a logo flourish?               |
| Production polish    | 0.15   | Lighting, motion, edit quality relative to the format's bar                    |

The other rubrics, at a glance:

| Format            | Dimensions (heaviest first)                                                                |
| ----------------- | ------------------------------------------------------------------------------------------ |
| `ugc_ad`          | First 3 seconds · native/un-ad feel · persona & trust · payoff/demo · CTA clarity          |
| `short_form_hook` | First 3 seconds · loop/retention · payoff · one clear idea                                 |
| `long_form_doc`   | Narrative arc · b-roll support · act breaks & re-hooks · retention engineering · authority |
| `explainer`       | Concept clarity · visual metaphor density · logical build · retention · hook               |
| `tutorial`        | Step clarity · completeness · pacing · hook/why-watch · screen clarity                     |
| `product_demo`    | Value clarity · demo flow · proof/outcome · CTA · production                               |
| `vlog`            | Hook · story through-line · personality · pacing & edit · payoff                           |
| `cinematic`       | Visual craft · mood/atmosphere · sound design · emotional arc                              |
| `news_commentary` | Argument clarity · hook/framing · credibility · pacing                                     |
| anything else     | Hook · structure · retention · clarity · production                                        |

Every dimension score comes with a `read` — a one-line, concrete note on *why* it scored that way, grounded in the blueprint and script data rather than generic praise.

## What to do with it

* **Read the `read` lines before the numbers.** "Hook 6.2 — opens on a tagline, tension arrives at second 9" tells you what to fix; the number alone doesn't.
* **Weight your revisions like the rubric weights.** On a B2B ad, a +1 on the hook moves the overall twice as much as a +1 on polish. Spend your iteration budget accordingly.
* **Compare videos within a format only.** A 7.4 tutorial and a 7.4 cinematic film are both good — at completely different things. Cross-format comparison is what the format-aware design exists to prevent.
* **If the format looks wrong, check the classification.** The scorecard inherits its rubric from [What it's selling](/layers/what-its-selling) — a misread format means a misfit rubric, and that layer records the evidence for its call.

## Related layers

<CardGroup cols={2}>
  <Card title="What it's selling" icon="bullseye" href="/layers/what-its-selling">
    The classification that picks this scorecard's rubric.
  </Card>

  <Card title="Hooks & re-hooks" icon="anchor" href="/layers/hooks">
    The hook dimension, unpacked.
  </Card>

  <Card title="Script analysis" icon="chart-line" href="/layers/script-analysis">
    The narrative data the scores are grounded in.
  </Card>

  <Card title="The Blueprint" icon="map" href="/layers/blueprint">
    The visual data behind the production dimensions.
  </Card>
</CardGroup>